Complement Activation by Whole Endotoxin Is Blocked by a ... · Sepsis is both a commonand, despite present-day therapy, a serious disease. ... This agent had no effect on the classical

INFECTION AND IMMUNITY, Oct. 1994, p. 4549-45550019-9567/94/$04.00+0Copyright © 1994, American Society for Microbiology

Complement Activation by Whole Endotoxin Is Blockedby a Monoclonal Antibody to Factor B

CHRISTOPHER W. CLARDY*

Department of Pediatrics, Rush Medical College, Chicago, Illinois 60612

Received 6 January 1994/Returned for modification 28 February 1994/Accepted 27 July 1994

Sepsis is both a common and, despite present-day therapy, a serious disease. The pathophysiology of theseptic response is a complex, multifactorial phenomenon which in part involves the activation of complementby bacterial endotoxin. A monoclonal antibody to human complement factor B, code-named lH5, which wascapable of specifically inhibiting the alternative pathway of complement activation at concentrations as low as1 ,ug/ml, is described. This agent had no effect on the classical pathway of complement activation. It wascapable of preventing the activation of complement by even high concentrations (0.1 mg/ml) of wholeendotoxin; however, it was ineffective in preventing activation of complement by endotoxin derived from arough mutant. This agent could potentially be used in the treatment of sepsis.

Sepsis is a systemic inflammatory response syndrome causedby the products of infection, particularly gram-negative bacte-ria. It is characterized initially by hyperdynamic shock resultingfrom a profound decrease in systemic vascular resistance.Subsequently, perturbations of the lungs, heart, kidneys, liver,and coagulation cascade can result in multiple organ failure.Sepsis is both a common and, despite present-day therapy, aserious disease, with 425,000 reported cases and approximately100,000 deaths in the United States last year (34).

It is generally acknowledged that the initial pathophysiologicevent in most cases of sepsis is the introduction of thelipopolysaccharide endotoxin from the outer cell walls ofgram-negative bacteria into the patient's circulation. Animals(22) and humans (47) inoculated with endotoxin develop ahyperdynamic state similar to early sepsis. In addition, 43% ofthe patients with clinical septic shock have detectable endo-toxemia (13).There are many endotoxin-induced mediators of sepsis. One

important mediator is the complement cascade, especially theanaphylatoxin C5a. The complement system may act alone orin conjunction with other mediators such as tumor necrosisfactor (24). There are three pieces of evidence that comple-ment activation participates in the pathophysiology of sepsis.First, complement activation occurs in sepsis. Bacteremicpatients have been shown to have decreased serum C3 con-centrations (15, 33) and increased serum C5a concentrations(6). In the latter case, the elevation of CSa correlated withclinical severity. Second, complement activation can mimicsepsis. Animals inoculated with homologous serum in whichthe complement had been activated by exposure to zymosandeveloped hyperdynamic shock similar to that found in septicpatients (37, 44). Finally, prevention of the formation of C5amitigates sepsis. Animals who either were congenitally defi-cient in C5 or were treated with antibody to CSa were lesslikely to develop hyperdynamic shock when inoculated withendotoxin than were normocomplementemic controls (24, 46).

Because of the apparent involvement of the complementcascade in endotoxin-mediated sepsis, a theoretic treatment ofsepsis would be the blockade of complement activation at a

* Corresponding author. Mailing address: Rush-Presbyterian-St.Luke's Medical Center, 1753 West Congress Pkwy., Chicago, IL 60612.Phone: (312) 942-4035. Fax: (312) 942-4168.

point prior to the formation of CSa. This could be done byblocking the activation of C3, the initial protein in the terminalpathway. However, this would deprive a patient of the ability tocombat a bacterial infection via the antibody-mediated classi-cal pathway of complement activation. Although it is contro-versial, there is evidence that complement activation by endo-toxin is mediated by the alternative pathway (15, 20, 21, 26).Therefore, blockade of this portion of the complement cascademay be sufficient to protect a patient from the deleteriouseffects of endotoxin-mediated complement activation. Thispaper describes a monoclonal antibody (MAb) to the humanform of the alternative-pathway protein complement factor B(FB), which, at pharmacologically significant concentrations, isable to completely block the activation of complement bywhole endotoxin.

(This work was presented at the Society of Critical CareMedicine's 1994 Annual Convention.)

MATERIALS AND METHODS

Reagents. Normal human serum (NHS) was obtained byvenipucture of healthy volunteers. Blood was allowed to clot atroom temperature in glass tubes. Serum was separated bycentrifugation, aliquotted, and stored at -70°C prior to use.Human FB was obtained from the Scripps Institute, La Jolla,Calif. There it was isolated from fresh human serum by saltingout the serum in ammonium sulfate and then subjecting it toDEAE- and CM-Sephadex chromatography (2). Polyacryl-amide gel electrophoresis of this preparation showed a singleband in the 90-kDa region. Sheep erythrocytes coated withantibody to hemolysin, goat anti-mouse immunoglobulin G(IgG)-peroxidase conjugate, mouse IgG standard, and zymo-san were obtained from Sigma Chemical Company, St. Louis,Mo. Endotoxin purified from Salmonella typhosa 0901 wasobtained from Difco Laboratories, Detroit, Mich. O-Polysac-charide-deficient endotoxin purified from Salmonella minne-sota R-595 was obtained from List Biological Laboratories,Minneapolis, Minn. Dulbecco's modified Eagle's medium(DMEM), equine serum, and fetal calf serum were obtainedfrom Hyclone Laboratories, Logan, Utah.

Production and characterization of MAbs directed againstFB (aFB MAbs). Four aFB MAbs were isolated in thefollowing manner. A female BALB/c mouse was injectedsubcutaneously with 30 ,ug of human FB, emulsified with an

4549

Vol. 62, No. 10

on February 4, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

4550 CLARDY

equal volume of Freund's complete adjuvant, every 4 weeks fora total of three injections. After this, preimmunization andpostimmunization serum samples were compared in order toconfirm the presence of antibodies directed against FB (seebelow for the antibody assay). The mouse was given anintraperitoneal injection of 30 ,ug of human FB in saline 4weeks after the final subcutaneous injection. Three days laterthe mouse was sacrificed and the spleen was aseptically re-moved.

Splenocytes were suspended in DMEM and fused with theimmortalized cell line P3-x63/Ag 8.653 (28) at a 3:1 ratio with45% polyethylene glycol by the technique of Galfre andMilstein (17). The fused cells were suspended in DMEMsupplemented with 10% (vol/vol) heat-inactivated fetal calfserum, 10% (vol/vol) heat-inactivated equine serum, hypoxan-thine, and thymidine, as well as aminopterin, which preventedthe growth of the thymidine kinase-negative myeloma cells,and incubated in tissue culture wells. After 10 days in ahumidified, 10% C02, 37°C atmosphere, the medium fromeach well was screened for antibodies to human FB (see belowfor the antibody assay) and positive wells were subcloned tomonoclonality. Working MAb stock was obtained by inoculat-ing large amounts of DMEM with otFB MAb-producing cellsand allowing these cells to incubate in a humidified, 10% C02,37°C atmosphere for approximately 10 days. The spent me-dium from these flasks was centrifuged at approximately 500 xg for 5 min to remove cellular debris. The IgG in this spentmedium was isolated from other confounding components byprecipitating the bulk of the non-IgG proteins with octanoicacid and then precipitating the IgG with ammonium sulfate(39). The ammonium sulfate-precipitated IgG was resus-pended in 0.01 M Tris HCl, pH 7.4, containing 0.14 M NaCl(TS), dialyzed versus TS, and centrifuged at 32,000 x g for 40min at 4°C to remove any aggregates. Polyacrylamide gelelectrophoresis of this preparation and Western blot (immu-noblot) analysis using a mouse IgG heavy and light chain-specific goat IgG conjugated with peroxidase (23, 32, 50)showed only those bands consistent with the heavy and lightchains of IgG.The titer of aoFB MAb was determined by the following

enzyme-linked immunosorbent assay (ELISA). Five micro-grams of human FB was placed in each well of a 96-wellpolystyrene plate and incubated overnight at 4°C. After beingwashed extensively with TS and blocked with bovine serumalbumin to prevent subsequent nonspecific protein binding,100 RI of a solution containing aFB MAb was added to eachwell and allowed to incubate. After further extensive washingof the wells, any u.FB MAb bound to the solid-phase FB wasdetected with goat anti-mouse IgG-peroxidase, which wassubsequently detected by its colorimetric reaction with o-phenylenediamine. To confirm whether a particular MAb wastruly directed against FB, a Western blot was done on adenatured polyacrylamide gel of purified FB as well as a bovineserum albumin control to confirm that the MAb bound only tothe 93,000-Da FB band (23, 32, 50).The avidity of each otFB MAb was determined by perform-

ing the above-described aFB MAb ELISA with 10-1 to 10-7M concentrations of mouse IgG (see below for the assay ofmouse IgG). A plot of the concentration of IgG ([Ab]) versusthe optical density (OD) produced a sigmoid curve which couldbe described by the equation

OD = maximum OD/[1 + exp(P3-y1n[Ab])] (1)

where maximum OD, 3, and -y were constants (40). The affinityconstant (K) for each oaFB MAb could be determined from this

curve by assuming that at the concentration of IgG whichproduced half the maximal OD ([Ab]50%), the amount ofbound antigen equalled the amount of unbound antigen.Therefore,

K = [Ag *Ab]/[Ag][Ab] = 1/[Ab]50% (2)

Combination of equations 1 and 2 yields the final equation

OD = maximum OD/{1 + exp[--yln(K[Ab])]} (3)which could be fitted to the [Ab] versus OD data by nonlinearregression analysis.The cxFB MAbs were isotyped by the following ELISA.

Sheep anti-mouse IgG (Boehringer Mannheim Corporation,Indianapolis, Ind.) was diluted in TS, placed in each well of a96-well polystyrene plate, and incubated overnight at 4°C.After being washed extensively and blocked with bovine serumalbumin to prevent nonspecific protein binding, 100 RI of anaFB MAb-containing solution was placed in each well andallowed to incubate. After further extensive washing of thewells, peroxidase-labelled, affinity-purified sheep antibodies tothe specific heavy- and light-chain isotypes of mouse immuno-globulins (Boehringer Mannheim Corporation) were placed inseparate wells such that each oLFB MAb was exposed to eachanti-isotypic antibody. After further incubation and washing ofthe wells, any bound peroxidase was detected as describedabove.Assay of murine IgG concentration. The concentrations of

murine IgG in the acFB MAb preparations were determined byELISA. Stock murine IgG at concentrations of 10 ng/ml to 10,ug/ml as well as samples of oxFB MAb preparations was dilutedin 0.1 M sodium carbonate, pH 9.6, to a total volume of 100 ,uland placed in wells of a 96-well polystyrene plate. The plateswere incubated overnight at 4°C. After being washed exten-sively with TS and blocked with bovine serum albumin toprevent subsequent nonspecific protein binding, 100 ,u of goatanti-mouse IgG-peroxidase was added to each well and al-lowed to incubate. The bound peroxidase was detected by itscolorimetric reaction with o-phenylenediamine. The aFB MAbpreparations contained 0.3 to 0.5 mg of murine IgG per ml.

Assays of complement activation. Complement activationwas determined by measuring the generation of C3a des arg byradioimmunoassay (Amersham Life Science Products, Arling-ton Heights, Ill.), by measuring the loss of the native antigen ofC3 by radial immunodiffusion as previously described (10), orby measuring the loss of the ability of serum to lyse antibody-coated sheep erythrocytes. In the latter method, serum sampleswere diluted in various amounts ofGVB+ + (141 mM NaCl-4.9mM sodium barbital [pH 7.4] with 0.15 mM CaCl2, 0.5 mMMgCl2, and 0.1% gelatin) and incubated at 37°C for 30 minwith 108 antibody-coated sheep erythrocytes per ml. Thereactions were stopped by the addition of EDTA to a finalconcentration of 50 mM. Cell suspensions were centrifuged,and the erythrocyte lysis was determined by measuring theabsorption of the supernatant at 412 nm. Data were fitted tothe Kabat-Meyer model by nonlinear regression analysis. He-molytic titers were expressed as reciprocals of the 50% hemo-lytic titer (31).When zymosan was used to activate complement in serum, it

was prepared as a suspension and vortexed immediately priorto addition. Zymosan-containing incubation mixtures werevortexed periodically to resuspend the zymosan. When endo-toxin was used as an activator, it was sonicated in a water bathat 37°C for 20 min prior to its addition. Mixtures containingendotoxin were continuously sonicated during incubation.The degree of inhibition of the alternative pathway pro-

INFECT. IMMUN.

on February 4, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

BLOCKADE OF ENDOTOXIN ACTIVATION OF COMPLEMENT 4551

duced by otFB MAb 1H5 was calculated by comparing theamount of C3a des arg produced in a reaction mixture withthat produced in the absence of otFB MAb 1H5 (zymosancontrol) and with that produced in fresh NHS (NHS control)by using the following formula:

% inhibition = 100 - (C3a des arg in sample - NHS control)/(zymosan control - NHS control).

Statistical analysis. Statistical analyses were done by usingthe two-tailed Student t test, analysis of variance with theFisher test for least significant difference, log-linear regressionanalysis (51), and nonlinear regression analysis using theGauss-Newton algorithm (54). Statistical significance is de-fined as a P value of '0.01.

RESULTSCharacterization of eFB MAbs. Four hybridoma cell lines,

code-named 1H5, 3F1, 5G7, and 7C3, were isolated. They allproduced IgG directed against FB [1H5, IgGl(K); 3F1,IgGl(K); 5G7, IgGl(X); and 7C3, IgG2a(K)]. All of the aFBMAbs had similar affinity constants (1H5, 4.0 x 109 ± 1.1 x109 M; 3F1, 2.5 x 109 ± 0.2 x 109 M; 5G7, 6.3 x 108± 1.0 x108 M; and 7C3, 2.9 x 108 ± 0.2 x 108 M).aFB MAbs did not cause complement activation. NHS was

mixed with equal volumes of each otFB MAb preparation (finalmurine IgG concentration of 150 ,ug/ml). These mixtures wereincubated at 37°C for 480 min. At 30-min intervals, 10-,ulaliquots were removed from the mixture and chelated by theaddition of EDTA to a final concentration of 56 mM in orderto stop all magnesium- and calcium-dependent complementactivation. The concentrations of native C3 (i.e., undegradedC3) in these aliquots as percentages of that found in undilutedNHS were determined by radial immunodiffusion as describedin Materials and Methods. The amount of C3 breakdown wasdetermined by comparing the native C3 concentrations in theinitial aliquots and subsequent aliquots. None of the oaFB MAbpreparations showed any ability to cause loss of C3. This isconsistent with these IgGs existing as monomers which wouldlack the multiple proximate Fc domains required for classical-pathway complement activation (30).aFB MAb 1H5 blocked alternative-pathway complement

activation. NHS was mixed with equal volumes of each aFBMAb preparation (final murine IgG concentration of 150,ug/ml) or of TS and incubated at 37°C for 20 min. Subse-quently, zymosan, an activator of the alternative pathway, wasadded to a final concentration of S mg/ml and the mixtureswere again incubated at 37°C for 30 min. The percent activa-tion of C3 was measured by assaying the loss of native C3 in thesamples incubated with zymosan compared with the loss inzymosan-free controls by radial immunodiffusion. Only one ofthe aFB MAbs, the one code-named 1H5, was able to inhibitthe activation of complement by zymosan (C3 loss: TS control,53.6% ± 2.2%; lH5, 2.4% ± 3.3%; 3F1, 54.4% ± 2.2%; 5G7,53.6% ± 2.2%; and 7C3, 54.4% ± 2.2%; n = 5 for all groups;analysis of variance with Fisher least significant difference test,P < 0.0001).

Subsequently, a similar incubation was performed withvarying concentrations of aFB MAb 1H5 as shown in Fig. 1.Inhibition of activation was first seen at an atFB MAb 1H5concentration of 1 ,ug/ml (third point in Fig. 1), and 100%inhibition was seen at approximately 25 ,ug/ml. These datawere fitted to the asymptotic model y = a(1 -e-bx) bynonlinear regression analysis, and the dose of aFB MAb 1H5required to produce 50% inhibition was found to be 6.11 ±1.20 ,ug/ml.

100-

80 -

(a cC.o0

C o

0 as

60 -

40-

.

20 -

0 25 50 75 100 125 150

[aFB MoAb 1 H5](RgImI)

FIG. 1. NHS was mixed with an equal volume of an otFB MAb 1H5preparation, diluted in TS to achieve the appropriate concentration ofthe MAb, and incubated at 37°C for 20 min. Subsequently, zymosanwas added to a final concentration of 5 mg/ml and the mixture wasagain incubated at 37°C for 30 min. The activation of the complementcascade was measured by assaying the production of C3a des arg by inthe NHS incubated with both cLFB MAb 1H5 and zymosan, in NHSincubated with zymosan alone (zymosan controls), and in fresh serum(NHS controls) by radioimmunoassay. The curve shown is the nonlin-ear regression analysis best fit of the asymptotic modely = a(1 -e-r)with a 50% effective dose of 6.11 ± 1.20 ,ug/ml (P < 0.005).

aFB MAb 1H5 had no eflect on classical-pathway comple-ment activation. NHS was mixed with an equal volume of apreparation of otFB MAb 1H5 (final murine IgG concentrationof 150 ,ug/ml) and incubated at 37°C for 20 min. The hemolytictiter of this NHS mixture, as well as that of control NHSincubated with TS, was measured as described in Materials andMethods. The hemolytic titer of the NHS-aFB MAb 1H5mixture was 78.43 ± 11.66, compared with a titer of 85.36 ±8.20 for the NHS-TS mixture (n = 8 for each group). Com-parison using a two-tailed Student t test showed that there wasno statistically significant difference between the two groups.

aLFB MAb 1H5 blockade of the alternative pathway pre-vented complement activation by whole endotoxin. NHS wasinitially incubated with aFB MAb 1H5, and subsequentlyvarious amounts of endotoxin derived from S. typhosa 0901were added. This bacterium's endotoxin contains both a lipidA-containing core lipopolysaccharide and an O-polysaccharidedomain, and it is therefore referred to as a whole endotoxin.The activation of the complement cascade was measured bythe production of C3a des arg. As can be seen in Fig. 2A, theactivation of the complement cascade, which occurred at 1 ,ugof added endotoxin per ml in the controls, was completelyprevented by preincubation with aFB MAb 1H5. Figure 2Bshows the same experiment done with endotoxin derived fromthe mutant S. minnesota R-595, which contains only the corelipopolysaccharide. Not only does aFB MAb 1H5 not preventthe activation of complement by this endotoxin, but it actuallyaccentuates the degree of activation seen in the control. This

VOL. 62, 1994

on February 4, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

4552 CLARDY

40000 l A 20000 -

18000 -

16000 -

0c

ca)-co g.C° 0o 14000 -

30000

CD,

co _

0(Dco gs

20000

10000 I0 T

12000 -

10000

i6:.i:. 0. 1. 210 -3 0 21o - 10 ° 10' 1 lo

[Endotoxin](jg/ml)

16 3 16 -2 10 -1 160° 10 1 10 2

[Endotoxin](n/ml)

FIG. 2. NHS was incubated with either aFB MAb 1H5 (final murine IgG concentration of 150 jig/ml) or TS for 20 min at 37°C, after whichvarious amounts of endotoxin were added. The NHS-aoFB MAb 1H5-endotoxin mixture was incubated for another 2 h at 37°C. The activation ofthe complement cascade was determined by assaying C3a des arg by radioimmunoassay. The samples incubated with otFB MAb 1H5 are shownas solid circles, and those incubated with TS are shown as open circles. Figure 2A shows the results of incubation with whole endotoxin derivedfrom S. typhosa 0901. Figure 2B shows the results of incubation with endotoxin derived from S. minnesota R-595. Note that the scale of they axisin Fig. 2A is different from that in 2B. Log-linear regression analysis of these data showed statistically significant complement activation withincreasing endotoxin levels in all samples except for the NHS-oLFB MAb 1H5-S. typhosa 0901 mixture.

results from the potentiation of the classical pathway notedwhen the alternative pathway of complement activation isblocked (20).

DISCUSSION

As detailed in the Introduction, activation of the comple-ment cascade, presumably by endotoxin, is involved in thepathophysiology of sepsis. This paper describes a MAb tohu'man FB, aFB MAb 1H5, which may be used to control thiscomplement activation.The complement cascade may be initiated via two pathways.

The classical pathway involves antigen-antibody complexes andthe complement proteins Clq, Clr, Cls, C2, and C4 to form aC3 convertase. The alternate pathway involves the comple-ment proteins C3b (or spontaneously hydrated C3), properdin,factor D, and FB to form a C3 convertase. The C3 convertasefrom either pathway will catalyze the cleavage of C3 andinitiate the subsequent activation of the terminal cascade. Asone would predict, the aFB MAb 1H5 described in this paperblocks only the alternate pathway of complement activation.The other three aFB MAbs described in this paper failed toblock the activation of the alternative pathway despite havingaffinities to human FB that are similar to that of crFB MAb

1H5. In three studies of ctFB MAbs, two MAbs which in-creased FB activity by stabilizing the alternative-pathway con-vertase (12), one which increased FB activity by enhancing thebinding of B to C3b (49), three which decreased FB activity bydestabilizing the alternative-pathway convertase (12, 52), andtwo which decreased FB activity by blocking the binding of FBto C3b (52) have been isolated. It is possible that oFB MAb1H5, as opposed to the other aoFB MAbs described in thispaper, bound to human FB in such a way as to block its bindingto C3b, so as to accelerate the decay of the alternative-pathwayconvertase or so as to deform its tertiary structure in such away as to make it inactive.The endotoxins found in the outermost cell wall of gram-

negative bacteria are complex molecules made up of a corelipopolysaccharide containing lipid A. This is covalently boundto a long, complex polysaccharide (0-polysaccharide). Theexact structure of the endotoxin varies a great deal from onebacterial species to another. In particular, the size of the0-polysaccharide can range from nonexistent to >106 Da. Ithas long been recognized that endotoxin is capable of activat-ing the complement cascade. It was initially shown that endo-toxin activated the alternative cascade (20), presumably be-cause the 0-polysaccharides prevent the inhibitory factor Hfrom having access to spontaneously formed alternative-path-

INFEcr. IMMUN.

on February 4, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

BLOCKADE OF ENDOTOXIN ACTIVATION OF COMPLEMENT 4553

way C3 convertase (38). It was subsequently shown that thelipid A portion of the endotoxin molecule was capable ofactivating the classical pathway of the complement cascade(35). At the present time there is not universal agreement as towhich of these pathways of complement activation is moresignificant in sepsis; however, there are several pieces ofevidence which indicate that the alternative pathway is moreimportant. First, in the previously cited study of complementlevels in septic patients (15), it was found that the decreasedserum C3 levels correlated with similar decreases in thealternative-pathway protein FB, whereas there was no detect-able decrease in any of the classical-pathway proteins. Second,in a series of studies by the same group (21, 26), a Salmonellastrain which could have the length of its O-polysaccharidevaried by manipulations of its growth medium was described.These bacteria were assayed for their ability to bind C3 and tobe killed in nonimmune serum. It was shown that C3 boundonly to bacteria containing endotoxin molecules with longO-polysaccharides and that this binding occurred even in thepresence of Mg2+-EDTA, which prevents classical-pathwayactivity. This is consistent with the theory that these bacteriaactivate complement only via the alternative pathway. Thisgroup further showed that when the lengths of the 0-polysac-charides on the bacteria were progressively diminished, thebacteria became more susceptible to being killed by nonim-mune serum. This would indicate that the ability of bacteria toactivate complement via the alternative pathway on their0-polysaccharides conferred resistance to the bacteria. Asimple explanation of this phenomenon would be that the longO-polysaccharides allow serum complement to be activated bythe alternative pathway in a location remote from the actualbacterial cell wall, thereby protecting the bacteria from themembrane attack complex formed during complement activa-tion. This would provide a teleologic explanation for theproduction of the O-polysaccharides by the bacteria and forthe alternative pathway activation seen during bacterial infec-tions.

Consistent with this interpretation, this paper shows thatblockade of the alternative pathway of complement activationby aFB MAb 1H5 prevents complement activation by wholeendotoxin but not by O-polysaccharide-deficient endotoxin.These results indicate that with whole endotoxin derived fromS. typhosa 0901, the alternative pathway is the exclusiveactivator of complement. If lipid A activation of the classicalpathway was occurring, then complement activation wouldproceed despite the block in the alternative pathway.

Because of the involvement of complement activation in thepathophysiology of a variety of diseases, many groups haveattempted to identify a "magic bullet" which would terminatethe turnover of the complement cascade without affectingother important biologic processes.

Initial efforts focused on cobra venom factor, a 140-kDaprotein derived from the venom of the cobra Naja naja. Thisprotein is homologous to the C3c portion of C3 (53), and, assuch, it binds to FB, is cleaved by complement factor D, andallows the formation of an active C3 convertase (36). Asopposed to the C3 convertase formed by true C3, the cobravenom factor C3 convertase is resistant to the actions of thecomplement control proteins H and I. This allows the undamp-ened in vitro and in vivo turnover of the complement cascadewith a resultant depletion of serum complement (11, 29).Cobra venom factor has never been attractive as a clinical tool,however, in large part because its mechanism of action, thepromotion of the uncontrolled turnover of the complementcascade, leads to the release of C3a and C5a into the circula-tion (8) and causes hematologic abnormalities such as hemo-

lysis, thrombolysis, and disseminated intravascular coagulation(5, 14).Attempts to inhibit the initiation of the complement cas-

cade, rather than to push it to exhaustion as with cobra venomfactor, have been made by producing substances which mim-icked the structure of the proteins involved in complementactivation to act as competitive inhibitors (3, 4, 7, 9, 18, 27, 42,43, 48). Although these attempts have met with some success,they have required very high concentrations of inhibitorysubstances ranging from 500 nM to 10,000,uM. As it wouldrequire gram amounts of these low-molecular-weight sub-stances to produce such concentrations in the human extracel-lular fluid, these compounds are unlikely to be pharmacolog-ically useful. The oFBMAb 1H5 described in this paper beginsto inhibit complement activation at concentrations as low as 1,ug/ml. This level is clinically achievable, as it is similar to theconcentrations of mouse IgG in serum seen with the antirejec-tion drug OKT3 (16). A potential limiting factor in thetreatment of humans with any mouse IgG would be thepresence of preformed human antibodies to mouse IgG, whichwould prevent the development of any therapeutic concentra-tion in serum. Although these antibodies are rarely seen inpatients who have not been previously treated with mouse IgG,they are seen in 20 to 85% of treated patients (19, 25). Adrawback to the treatment of sepsis with aFB MAb 1H5 wouldbe the development of these blocking antibodies, which wouldlimit further treatment with other mouse monoclonal agents.

Recently, a more successful attempt to obtain an effectiveinhibitor of complement activation was described by Weismanand coworkers (55). They exploited an analog of the comple-ment control protein complement receptor 1, a membraneprotein which allows cells to bind to the activated complementproteins C3b and C4b (1, 41). This analog, sCR1, was able toblock the activation of both the classical and the alternatecomplement pathways in vitro at doses as low as 10 to 20,ug/ml, similar to the effective dose of aFB MAb 1H5. Atheoretic disadvantage of sCR1 compared with aFB MAb 1H5is that it does not selectively block the alternative pathway ofthe complement cascade. Although blockade of the classicalpathway may be potentially advantageous in some clinicalsettings, in the treatment of sepsis it would have the drawbackof hindering the host's ability to combat bacterial invasion byusing specific antibacterial antibodies. An agent that blockedonly the alternative pathway would prevent formation of C5aby endotoxin but would leave other complement-mediateddefenses intact. It must be noted that blockade of the alterna-tive pathway itself may interfere with the host's ability todefend itself from bacterial attack. Patients who have congen-ital deficiencies of the alternative pathway protein properdinhave an increased incidence of meningococcal disease (45).Further, there are no descriptions of persons with FB deficien-cies, which may indicate that blockade of this protein hasprofoundly deleterious consequences.

In summary, this paper describes the derivation of aFBMAb 1H5 and demonstrates that it is a potent and specificinhibitor of the alternative pathway of complement activation.It was further shown that by inhibiting the alternative pathway,aFB 1H5 is able to prevent the activation of complement bywhole endotoxin. This agent has theoretic potential in thetreatment of gram-negative sepsis. Future work will test theability of aFB MAb 1H5 to blunt the cytokine response toendotoxin of immunopotent cells in tissue culture as well as itsability to mitigate the septic response in animal models.

VOL. 62, 1994

on February 4, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

4554 CLARDY

REFERENCES1. Ahearn, J. M., and D. T. Fearon. 1989. Structure and function of

the complement receptors, CR1 (CD35) and CR2 (CD21). Adv.Immunol. 46:183-219.

2. Alper, T., and C. A. Alper. 1970. Isolation and properties of aglycine-rich ,B-glycoprotein of human serum. Biochim. Biophys.Acta 221:529-539.

3. Andreatta, R. H., J. Rahuel, M. Wesp, and P. Dukor. 1981.Attempts at specific inhibition of C3 convertase. In U. Brodbeck(ed.), Enzyme inhibitors. Verlag Chemie, Basel.

4. Asghar, S. S. 1984. Pharmacologic manipulation of complementsystem. Pharmacol. Rev. 36:223-244.

5. Ballow, M., and C. G. Cochrane. 1969. Two anticomplementaryfactors in cobra venom: hemolysis of guinea pig erthrocytes by oneof them. J. Immunol. 103:944-952.

6. Bengtson, A., and M. Heideman. 1988. Anaphylatoxin formation insepsis. Arch. Surg. 123:645-649.

7. Boackle, R. J., B. J. Johnson, and G. B. Caughman. 1979. An IgGprimary sequence exposure theory for complement activationusing synthetic peptides. Nature (London) 282:742-743.

8. Bokisch, V. A., H. J. Muller-Eberhard, and C. G. Cochrane. 1969.Isolation of a fragment (C3a) of the third component of humancomplement containing anaphylatoxin and chemotactic activityand description of an anaphylatoxin inactivator of human serum. J.Exp. Med. 129:1109-1130.

9. Caporale, L. H., S. S. Gaber, W. Kell, and 0. Gotze. 1981. Afluorescent assay for complement activation. J. Immunol. 126:1963-1965.

10. Clardy, C. W., J. Forristal, C. F. Strife, and C. D. West. 1989. Aproperdin dependent nephritic factor slowly activating C3, C5, andC9 in membranoproliferative glomerulonephritis, types I and III.Clin. Immunol. Immunopathol. 50:333-347.

11. Cochrane, C. G., H. J. Muller-Eberhard, and B. S. Aikin. 1970.Depletion of plasma complement in vivo by a protein of cobravenom: its effect on various immunologic reactions. J. Immunol.105:55-69.

12. Daha, M. R., A. M. Deeler, and K. A. Van Es. 1984. Stabilizationof the amplification convertase of complement by monoclonalantibodies directed against human factor B. J. Immunol. 132:2538-2542.

13. Danner, R. L., R. J. Elin, J. M. Hosseini, R. A. Wesley, J. M. Reilly,and J. E. Perillo. 1991. Endotoxemia in human septic shock. Chest99:169-175.

14. Dodds, W. J., and R. J. Pickering. 1972. The effect of cobra venomon hemostasis in guinea pigs. Blood 40:400-411.

15. Fearon, D. T., S. Ruddy, P. H. Schur, and W. R. McCabe.1975. Activation of the properdin pathway of complement inpatients with gram-negative bacteremia. N. Engl. J. Med. 292:937-940.

16. First, M. R., T. J. Schroeder, P. E. Hurtubise, M. E. Mansour, L.Penn, R. Munda, W. F. Balistreri, J. W. Alexander, D. B. Melvin,J. P. Fidler, F. C. Ryckman, and M. E. Brunson. 1989. Successfulretreatment of allograft rejection with OKT3. Transplantation47:88-91.

17. Galfre, G., and C. Milstein. 1981. Preparation of monoclonalantibodies: strategies and procedures. Methods Enzymol. 73:3-46.

18. Glover, G. I., C. S. Schasteen, W. S. Liu, and R. P. Levine. 1988.Synthetic peptide inhibitors of complement serine proteases. I.Identification of functionally equivalent protease inhibitor se-quences in serpins and inhibition of Cls and D. Mol. Immunol.25:1261-1267.

19. Goldstein, G., A. J. Fucello, D. J. Norman, et al. 1986. OKT3monoclonal antibody plasma levels during therapy and the subse-quent development of host antibodies to OKT3. Transplantation42:507-511.

20. Gotze, O., and H. J. Muller-Eberhard. 1971. The C3-activatorsystem: an alternate pathway of complement activation. J. Exp.Med. 134:90s-108s.

21. Grossman, N., M. A. Schmetz, J. Foulds, E. N. Klima, V. Jimenez,L. L. Leive, and K. A. Joiner. 1987. Lipopolysaccharide size anddistribution determine serum resistance in Salmonella montevideo.J. Bacteriol. 169:856-863.

22. Guenter, C. A., V. Fiorica, and L. B. Hinshaw. 1969. Cardiorespi-

ratory and metabolic responses to live E. coli and endotoxin in themonkey. J. Appl. Physiol. 26:780-786.

23. Harlow, E., and D. Lane. 1988. Antibodies, a laboratory manual, p.471-511. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y.

24. Hsueh, W., X. Sun, L. N. Rioja, and F. Gonzalez-Crussi. 1990. Therole of the complement system in shock and tissue injury inducedby tumor necrosis factor and endotoxin. Immunology 70:309-314.

25. Jaffers, G. J., T. C. Fuller, A. B. Cosimi, et al. 1986. Monoclonalantibody therapy: anti-idiotypic and non-idiotypic antibodies toOKT3 arise despite intense immunosuppression. Transplantation41:572-578.

26. Joiner, K. A., N. Grossman, M. Schmetz, and L. Leive. 1986. C3binds preferentially to long-chain lipopolysaccharides during al-ternative pathway activation by Salmonella montevideo. J. Immu-nol. 136:710-715.

27. Kaneko, I., D. T. Fearon, and K. F. Austen. 1980. Inhibition of thealternative pathway of human complement in vitro by a naturalmicrobial product, complestatin. J. Immunol. 124:1194-1198.

28. Kearney, J. F., A. Radbruch, B. Liesegang, and K. A. Rajewsky.1979. A new mouse myeloma cell line that has lost immunoglob-ulin expression but permits the construction of antibody-secretinghybrid cell lines. J. Immunol. 123:1548-1550.

29. Klein, P. G., and J. Wellensiek. 1965. Multiple nature of the thirdcomponent of guinea-pig complement. I. Separation and charac-terization of three factors a, b and c, essential for hemolysis.Immunology 8:590-603.

30. Lachman, P. J., and N. C. Hughes-Jones. 1984. Initiation ofcomplement activation. In H. J. Muller-Eberhard and P. A.Miescher (ed.), Complement. Springer Verlag, Berlin.

31. Lachmann, P. J., M. J. Hobart, and W. P. Aston. 1973. Comple-ment technology. In D. M. Weir (ed.), Handbook of experimentalimmunology, 2nd ed. Blackwell Scientific Publications, Oxford.

32. Laemmli, U. K. 1970. Cleavage of structural proteins during theassembly of the head of bacteriophage T4. Nature (London)227:680-685.

33. McCabe, W. R. 1973. Serum complement levels in bacteremia dueto gram-negative organisms. N. Engl. J. Med. 288:21-23.

34. Morbidity and Mortality Weekly Report. 1990. Increase in na-tional hospital discharge survey rates for septicemia-UnitedStates, 1979-1987. Morbid. Mortal. Weekly Rep. 39:31-34.

35. Morrison, D. C., and L. F. Kline. 1977. Activation of the classicaland properdin pathways of complement by bacterial lipopolysac-charides (LPS). J. Immunol. 118:362-368.

36. Miiller-Eberhard, H. J. 1967. Mechanism of inactivation of thethird component of human complement (C'3) by cobra venom.Fed. Proc. 26:744.

37. Nuytinck, J. K. S., R. J. A. Goris, J. G. E. Weerts, P. H. M.Schillings, and J. H. Schuurmans-Stekhoven. 1986. Acute gener-alized microvascular injury by activated complement and hypoxia:the basis of the adult respiratory distress syndrome and multipleorgan failure? Br. J. Exp. Pathol. 67:537-548.

38. Pangburn, M. K. 1989. Analysis and recognition in the alternativepathway of complement: effect of polysaccharide size. J. Immunol.142:2766-2770.

39. Perosa, F., R. Carbone, S. Ferrone, and F. Dammacco. 1990.Purification of human immunoglobulins by sequential precipita-tion with caprylic acid and ammonium sulphate. J. Immunol.Methods 128:9-16.

40. Ratkowsky, D. A. 1986. Choosing near-linear parameters in thefour-parameter logistic model for radioligand and related assays.Biometrics 42:575-582.

41. Ross, G. D., and M. E. Medof. 1985. Membrane receptors specificfor bound fragments of C3. Adv. Immunol. 37:217-267.

42. Schasteen, C. S., R. P. Levine, S. A. McLafferty, R. F. Finn, L. D.Bullock, J. C. Mayden, and G. I. Glover. 1991. Synthetic peptideinhibitors of complement serine proteases. III. Significant increasein inhibitor potency provides support for the functional equiva-lence hypothesis. Mol. Immunol. 28:17-26.

43. Schasteen, C. S., S. A. McLafferty, G. I. Glover, C. Y. Han, J. C.Mayden, W. S. Liu, and R. P. Levine. 1988. Synthetic peptideinhibitors of complement serine proteases. II. Effects on hemolyticactivity and production of C3a and C4a. Mol. Immunol. 25:1269-1275.

INFECT. IMMUN.

on February 4, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

BLOCKADE OF ENDOTOXIN ACTIVATION OF COMPLEMENT 4555

44. Schirmer, W. J., J. M. Schirmer, G. B. Naff, and D. E. Fry. 1988.Systemic complement activation produces hemodynamic changescharacteristic of sepsis. Arch. Surg. 123:316-321.

45. Sjoholm, A. G., J. H. Praconier, and C. Soderstrom. 1982.Properdin deficiency in a family with fulminant meningococcalinfections. Clin. Exp. Immunol. 50:291-297.

46. Smedegard, G., L. Cui, and T. E. Hugli. 1989. Endotoxin inducedshock in the rat-a role for C5a. Am. J. Pathol. 135:489-497.

47. Suffredini, A. F., R. E. Fromm, M. M. Parker, M. Brenner, J. A.Kovacs, R. A. Wesley, and J. E. Perillo. 1989. The cardiovascularresponse of normal humans to the administration of endotoxin. N.Engl. J. Med. 321:280-287.

48. Takada, Y., Y. Arimoto, H. Mineda, and A. Takada. 1978. Inhibi-tion of the classical and alternative pathways by amino acids andtheir derivatives. Immunology 34:509-515.

49. Tanaka, E., K. Hong, T. Kinoshita, Y. Takata, H. Kozono, J.Takeda, A. Yoden, and K. Inoue. 1991. Murine monoclonalanti-Ba antibody that enhances haemolytic activity of factor B.Immunology 73:383-387.

50. Towbin, H., T. Staehelin, and J. Gordon. 1979. Electrophoretic

transfer of proteins from polyacrylamide gels to nitrocellulosesheets: procedure and some applications. Proc. Natl. Acad. Sci.USA 76:4350-4354.

51. Udnyyule, G., and M. G. Kendall. 1948. An introduction to thetheory of statistics. C. Griffin and Co. Ltd., London.

52. Ueda, A., J. F. Kearney, K. H. Roux, and J. E. Volanakis. 1987.Probing functional sites on complement protein B with monoclo-nal antibodies: evidence for C3b-binding sites on Ba. J. Immunol.138:1143-1149.

53. Vogel, C. W., C. A. Smith, and H. J. Muller-Eberhard. 1984. Cobravenom factor: structural homology with the third component ofhuman complement. J. Immunol. 133:3235-3241.

54. Watts, D. G., and D. M. Bates. 1985. Nonlinear regression analysis.In S. Kotz and W. W. Johnson (ed.), Encyclopedia of statisticalsciences. Wiley and Sons, New York.

55. Weisman, H. F., T. Bartow, M. Leppo, H. C. Marsh, G. R Carson,M. F. Concino, M. P. Boyle, K. H. Roux, M. L. Weisfeldt, and D. T.Fearon. 1990. Soluble human complement receptor type 1: in vivoinhibitor of complement suppressing post-ischemic myocardialinflammation and necrosis. Science 249:146-151.

VOL. 62, 1994

on February 4, 2020 by guest

http://iai.asm.org/

Dow

nloaded from